TY - GEN
T1 - Capturing Performance Requirements of Real-Time Systems Using UML/MARTE Profile
AU - Khakhar, Disha
AU - Nayak, Ashalatha
PY - 2018/1/1
Y1 - 2018/1/1
N2 - Performance is a critical parameter for successful development of real-time systems since their correctness is based not only on logical behavior, but also on timeliness of output. Traditional software development methods use ‘fix it later’ approach where focus is on correctness of the software and performance considerations are made in the later phases of software development. If performance problems are discovered at this point, software is modified to fix performance issues. This technique does not work well for time critical systems. In order to address this problem, Model Driven Software Performance Engineering (MDSPE) approach is used to include performance analysis in the early stage of software development life cycle. Performance parameters are associated with UML model elements using UML profile for MARTE, to capture software requirement in the design phase. Annotated UML model is transformed to various performance model in order to perform analysis. By evaluating the performance model, it is possible to obtain various performance output parameters using simulation and analytical techniques. These parameters will help in evaluating alternative design for the same system. Currently, there are no approaches that investigate the issue of annotation of existing UML diagrams with MARTE profile. The proposed research focuses on capturing performance requirements by annotation of UML models using UML profile for MARTE.
AB - Performance is a critical parameter for successful development of real-time systems since their correctness is based not only on logical behavior, but also on timeliness of output. Traditional software development methods use ‘fix it later’ approach where focus is on correctness of the software and performance considerations are made in the later phases of software development. If performance problems are discovered at this point, software is modified to fix performance issues. This technique does not work well for time critical systems. In order to address this problem, Model Driven Software Performance Engineering (MDSPE) approach is used to include performance analysis in the early stage of software development life cycle. Performance parameters are associated with UML model elements using UML profile for MARTE, to capture software requirement in the design phase. Annotated UML model is transformed to various performance model in order to perform analysis. By evaluating the performance model, it is possible to obtain various performance output parameters using simulation and analytical techniques. These parameters will help in evaluating alternative design for the same system. Currently, there are no approaches that investigate the issue of annotation of existing UML diagrams with MARTE profile. The proposed research focuses on capturing performance requirements by annotation of UML models using UML profile for MARTE.
UR - http://www.scopus.com/inward/record.url?scp=85036453210&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85036453210&partnerID=8YFLogxK
U2 - 10.1007/978-981-10-5687-1_62
DO - 10.1007/978-981-10-5687-1_62
M3 - Conference contribution
AN - SCOPUS:85036453210
SN - 9789811056864
T3 - Advances in Intelligent Systems and Computing
SP - 703
EP - 714
BT - Soft Computing
PB - Springer Verlag
T2 - 1st International Conference on Soft Computing: Theories and Applications, SoCTA 2016
Y2 - 28 December 2016 through 30 December 2016
ER -